search

NOAA-EMC / EMC_verif-global

Global Forecast System (GFS) verification package using MET and METplus
8 stars 14 forks source link

Update Jet support #110

Closed DavidHuber-NOAA closed 1 year ago

DavidHuber-NOAA commented 1 year ago

A Jet module file should be created and all components re-tested to verify they still work on the system. Similarly, the modules should all be pointed towards the EPIC hpc-stack builds. First, MET and METplus need to be installed in that hpc-stack distribution.

malloryprow commented 1 year ago

@DavidHuber-NOAA Is there are plan for MET and METplus to be installed in the hpc-stack?

DavidHuber-NOAA commented 1 year ago

@malloryprow It is available to be installed and newer versions of MET/METplus (10.1.2/4.1.3) have already been installed on Hera and Orion (I believe to support UFS SRW validation). For Jet, it's desired because the current installations in /contrib use older dependencies, making it difficult to play nicely with the workflow.

malloryprow commented 1 year ago

Are MET v9.1.3 and METplus v3.1.1 installed? EMC_verif-global uses those versions. I haven't updated to the most recent version because of the development for https://github.com/NOAA-EMC/EVS I have been doing the past year+.

DavidHuber-NOAA commented 1 year ago

@malloryprow Those versions aren't installed yet, but I have requested them. And thank you for the info RE EVS! That's very good to know that a followup system is in the works.

DavidHuber-NOAA commented 1 year ago

EPIC was able to install MET/METplus 9.1.3/3.1.1 on Jet, which I was able to cycle test. PR forthcoming.

malloryprow commented 1 year ago

Great, I'm heading out for the day soon, but will check it out tomorrow morning. Though I have on my calendar there is a Jet Outage tomorrow? Not sure if you want me to check any output, or if you feel the output from https://github.com/NOAA-EMC/global-workflow/pull/1301 suffices.

DavidHuber-NOAA commented 1 year ago

The logs do have a few messages in them that might be worth checking out, though they don't seem to affect the outcomes. I've pushed the logs over to Hera (/scratch1/NESDIS/nesdis-rdo2/David.Huber/para/com/jet_tc_192_logs/*/gfsmetp*). It looks like there may be a bug in run_verif_global_in_global_workflow.sh. There are a few places (e.g. line 145) where SDATE_GFS_YYYYMMDDHH is not in the expected format (2022-04-01 instead of 2022040100, for instance). I suspect this is a bug on all platforms, not just Jet.